Although the public debate on poverty issues is frequently emotional. Why So Many Children Are Poor. David M. Betson Robert T.

Transcription

1 25 Why So Many Children Are Poor David M. Betson Robert T. Michael Abstract According to the official U.S. measure of poverty, in 1995 the child poverty rate in this country was nearly 21%, compared with an adult poverty rate of 11%. This article explores why, according to the official measure, there are so many poor children. Working from the premise that children are poor because they live with poor adults, the reasons for adult poverty are reviewed. Both economic forces and demographic trends have contributed to growing inequality of earnings among workers. That inequality coupled with stagnating real earnings has increased poverty. In addition, education, age, and race affect an individual s earning capacity; the article examines the likelihood that an individual will earn enough to keep his or her family out of poverty, given the individual s educational attainment, age, and race. The reasons for the large difference between the child and adult poverty rates are explored, using a decomposition of the poverty population to show how demographic characteristics such as higher fertility rates among poor families and the higher prevalence of singleparent families among the poor lead to substantially higher poverty rates for children than for adults. Finally, the article examines the validity of the official poverty measure and reviews how an alternative measure proposed by a National Research Council panel would address the official measure s shortcomings. If the panel s proposed measure were adopted, it would change the statistical face of poor children. It would, for example, show an increase in the proportion of poor children who live in families with two parents and a corresponding decrease in the proportion in families with only one parent, and it would show an increase in the proportion of children who live in families with at least one full-time employed adult and a corresponding decrease in the proportion in families with no adult employed full time. David M. Betson, Ph.D., is an associate professor of economics at the University of Notre Dame. Robert T. Michael, Ph.D., is the Eliakim Hastings Moore Distinguished Service Professor at the Irving B. Harris Graduate School of Public Policy Studies at the University of Chicago. Although the public debate on poverty issues is frequently emotional and acrimonious, it is clear that an informed debate over policies to prevent or ameliorate the consequences of poverty for children requires an understanding of how poverty is measured, of the composition of the population of poor children, of the reasons for childhood poverty, and of the ability of public and private programs to address the needs of poor children. These issues are the focus of this issue of The Future of Children. To help frame the discussion of developing effective child-focused poverty policies, this article describes how poverty is officially defined in the United States, reviews recent trends in poverty according to that definition, The Future of Children CHILDREN AND POVERTY Vol. 7 No. 2 Summer/Fall 1997

2 26 THE FUTURE OF CHILDREN SUMMER/FALL 1997 and discusses the reasons for the widening gap between child and adult poverty rates. In addition, it also derives implications for policy and discusses how the official definition of poverty might be improved to reflect more accurately the real resources available to children living at the lower end of the economic spectrum. The concept of poverty is simple and familiar: a child living in poverty lacks goods and services considered essential to human well-being. However, devising an operational measure of poverty that accurately reflects a child s well-being in terms of his or her economic resources is complex. A child s well-being depends both on how many resources are available to his or her family and on how the adults in the family allocate those resources among family members. Resources include the goods and services purchased for the child by the family, those provided by the government, and those produced within the family. In the United States, families and individuals are classified as being in poverty if their annual money income (including cash earnings, unemployment insurance benefits, cash benefits from other government programs, and other sources of regular nonearnings income) before taxes and other deductions falls below official poverty thresholds. These thresholds were developed in the mid-1960s; they are adjusted annually for inflation. One set of thresholds reflects the needs of families by size and composition. For example, the 1995 poverty threshold for a single individual was $7,929 while for a family of four (two adults and two children), the threshold was $15,455, and for a family of six (with two adults and four children), the threshold was $20,364. The trend in poverty for all adults (age 18 and over) and all children in the United States over the past three decades is presented in Figure 1. Four characteristics of child poverty stand out: (1) the rate of poverty for children is currently around one in five children; (2) since the late 1950s, there has been a consistently higher rate of poverty for children than for adults; (3) after declining steadily from 1959 through 1969, the poverty rate for children increased by more than 50% by 1983, and the gap between the poverty rates for children and for adults also increased substantially during the same period; and (4) for the past decade, the poverty rate for children has fluctuated around 21%. 1 The first section of this article discusses why child poverty is currently so high relative to adult poverty. The second section discusses the trends seen in Figure 1. As discussed in the spring 1993 issue of The Future of Children, the methods used by the federal government to measure poverty have been criticized almost since the government adopted its official measure in The National Research Council (NRC) of the National Academy of Sciences has recently recommended a major revision of the measure. While the official measure indicates the extent and character of poverty among children, it is flawed because it does not reflect the impact of many existing federal welfare programs such as food stamps, public housing, school lunch, energy assistance, or Medicaid. The third section describes these flaws and the NRC s recommended new measure of poverty and indicates how this new measure would alter the picture of which children are poor.

3 Why So Many Children Are Poor 27 Although flawed, the official poverty measure is important because it is the reference point for discussions of and research on poverty, including the papers in this issue. Moreover, the transition from the old measure of poverty to a new measure has not yet taken place. Thus, for setting a frame of reference, the official measure is used in the first three sections of this article. Recent events increase the need for the adoption of a new measure of poverty. The enactment of the 1996 welfare reform law, the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) of 1996, shifts much of the responsibility for those with inadequate incomes from the federal government to state governments. As states adopt different strategies in response to their new responsibilities, the measurement of child poverty will become increasingly important. It will be essential to learn from the states efforts which programs work well and which ones work poorly. But to learn from these diverse state initiatives, a clear, internally consistent, and comprehensive measure of poverty will be needed. Thus, the need for a better measure has become more urgent. The NRC s recommendations provide such a measure of poverty. While the focus of this article and of most policy-oriented discussions of poverty is on the official poverty measure, it is worth noting that some policy issues cannot be addressed by a metric that simply classifies families and individuals as poor or nonpoor. For example, the differences in incomes among poor families are not reflected in the simple count of those who are poor; families that have resources well below the poverty threshold present different circumstances and different policy challenges than ones with incomes just below the poverty threshold. Also, although the focus here is on the children who are below a poverty threshold, there is, of course, much inequality of income among the four out of five children who are not poor. Focusing exclusively on poor children ignores the inequality among children in middle- and upper-income families. Finally, looking at which children live in families above or below a poverty threshold at one point in time tells us little about the lifetime or long-term experience of those counted as poor. The consequences of the duration of poverty suffered by a child are discussed in the articles by Corcoran and Chaudry and by Brooks-Gunn and Duncan in this journal issue. Reasons for Child Poverty The major question of this article is why, according to the official measure of poverty, more than 20% of the nation s 67 million children are poor. There is no single answer to this simple, but perplexing, question. Children live with adults and rely upon those adults for their economic well-being. By and large then, children are poor because they live with adults who are poor. Thus, to understand child poverty one must look to the causes underlying adult poverty, such as economic and demographic forces and factors affecting individual earning capacity. The rate of poverty for children is almost double that for adults, as shown in Figure 1, so it is necessary to examine why children are more likely than adults to reside in poor families. A decomposition of the population of poor families by demographic characteristics, such as number of adults and number of children per family, sheds some light on this question. Children Born to Poor Adults When considering the ability of adults to meet the needs of their families, a distinction is made among three groups: adults who are family self-sufficient, those who are adult self-sufficient, and those who do not even

4 28 THE FUTURE OF CHILDREN SUMMER/FALL 1997 Figure 1 Child and Adult Poverty Rates, 1959 to Percentage Poor by Official Measure Year Child Poverty Rate Adult Poverty Rate Source: U.S. Bureau of the Census. Poverty in the United States: Current Population Reports, P-60, no Washington, DC: U.S. Government Printing Office, September attain adult self-sufficiency. Adults are defined as family self-sufficient if their own resources (their income and their receipts from social insurance programs that are not dependent on the presence of children) are enough to keep their entire family out of poverty. Self-sufficient adults are defined as those who would have enough resources to maintain themselves above the poverty threshold if their children were not present but do not have enough to meet the additional needs of their children. This difference is reflected in the poverty thresholds for families of varying sizes. In 1995, the threshold was $10,205 for two nonelderly adults, and it was $15,455 if two children were included in this family. Using this terminology, an adult couple with an income of $12,000 in 1995 and two children would be classified as adult self-sufficient but not family self-sufficient because they would have had sufficient resources to exceed the official poverty threshold for two adults but not enough to meet the poverty standard for a family of two adults and two children. Clearly, the distinction being drawn here is only a first approximation, but it is a useful one. 3 Using data from the March 1993 Current Population Survey (CPS), a monthly survey of about 60,000 households in the United States carried out by the Census Bureau, it is possible to determine the percentage of adults who live in families at each of these three levels of sufficiency. 4 In 1992, some 10.2% of adults in families with children lacked adult self-sufficiency. Because 10.3% of adults without children are also classified as poor, it appears that adults who live with children are not more likely to be poor than a random sample of all adults. In addition, 5.5% of all adults with children were adult self-sufficient but failed to achieve family self-sufficiency. Based solely on the resources brought to the family by the adults, therefore, a total of 15.7% (= ) of all adults with children did not meet the needs of their families and would have been counted as poor in 1992 were it not for government cash welfare programs designed for children. The impact of these cash programs is discussed below. Why Are These Adults Poor? The determinants of adult poverty can be classified into two general categories:

5 Why So Many Children Are Poor 29 (1) macroeconomic and demographic forces which affect the overall income distribution and (2) factors that affect an individual s earning capacity, such as education, age, and race. 5 Economic and demographic forces. Over the past three decades, there has been an increase in the inequality of earnings among workers, resulting in a larger proportion of the population in poverty. The rise in inequality might not have led to higher poverty rates among adults if it had been accompanied by a substantial rise in the average level of real (inflation-adjusted) earnings, but in these past two decades, average real earnings have not grown much. Traditionally, unemployment has been a major factor influencing the poverty rate, 6 but despite the fact that unemployment has fallen by nearly one-third since 1983, the low and declining wages of less-educated workers have led to greater poverty. The major demographic trends of the past several decades are dramatic growth in the proportion of women who are in the labor force and the entry of those born during the baby boom. Together, these two groups contribute to a younger and more female-intensive workforce. The young have less job experience and, thus, receive lower wages, and females traditionally have experienced lower earnings, so both of these demographic factors have tended to create greater inequality in earnings. With regard to family income, however, the contributions of women s earnings to two-income families may have partly offset the trend resulting from greater inequality of individual earnings. Individual earnings capacity. Education, race, and age are among the personal factors that most affect earnings and income; they play a role in determining which and how many families are poor. Education level is an indicator of market skills that yield higher earnings; age proxies for job skills acquired through experience; and race, because of discrimination, affects both job market opportunities and payoffs. Table 1 demonstrates the association between each of these characteristics and an adult s ability to attain adult self-sufficiency. The education of the adults in the family is critical for the family income. The proportion of adults who are not in the labor force and the proportion who experience at least one period of unemployment during the year both decline dramatically as education rises. For example, the proportion of high school graduates who are not adult selfsufficient (10%) is less than half the proportion of nongraduates who are not adult selfsufficient (24%). The education panel of Table 1 shows that the distribution of adults and of children by the education level of the primary person in the family are, surprisingly, almost identical. 7 The age panel reminds us that children live with relatively young adults who have not Education, race, and age are among the personal factors that most affect earnings and income; they play a role in determining which and how many families are poor. typically reached their peak earning capacity and are (except for the retired elderly) relatively less likely to be adult self-sufficient; this is especially true for those under age 25, which helps explain in part the social concern about teenagers having children. Race also has a strong relationship to poverty. As compared with whites, black adults have lower earnings on average and are more likely to be unemployed or out of the labor force. The race panel of Table 1 shows that black adults are three times as likely as white adults to have incomes that are too low to meet even the adult s needs in the family. While each factor education, age, and race separately can greatly influence an adult s chances of being able to support the adults in the family, let alone the children as well, consider the chances of a young, black, high school dropout. An adult with these three attributes has a 15% chance of being adult self-sufficient. However, a young black who graduates from high school has a much higher (42%) chance of being adult self-sufficient. 4 The Role of Federal Government Cash Assistance As discussed above, in 1992, a total of 15.7% of adults who lived with children did

6 30 THE FUTURE OF CHILDREN SUMMER/FALL 1997 Table 1 Distribution of Adults and Children by Characteristics of the Primary Person in the Family Adults Percentage Characteristics Adults Children Percentage Not Adult Average Out of Percentage Self-Sufficient c Earnings Labor Force a Unemployed b All Adults $23, Education Less than high school , High school , Some college , Bachelor s degree , Advanced degree , Age Less than , to , to , to , to , to , to , to , to , and older , Race White , Black , Ethnicity Hispanic d , a b c d Out of labor force means not employed and not currently seeking a job. Unemployed means not employed and currently seeking a job. Adult self-sufficiency is defined as having sufficient resources to maintain the adults in a family above the official poverty level. Hispanic may include persons of any race. Source: Calculations by the authors from the March 1993 Current Population Survey. See Current Population Survey [CD-ROM], U.S. Bureau of the Census, Available from U.S. Department of Commerce, Bureau of the Census, P.O. Box , Atlanta, GA not attain family self-sufficiency. Of these 15.7% of adults, only 1.3% were raised out of poverty by cash assistance from federal programs such as Aid to Families with Dependent Children (AFDC, welfare payments to low-income families with children) and Supplemental Security Income (SSI, payments to low-income persons of all ages with disabilities). Thus, even with cash assistance, 14.4% of all adults with children in 1992 remained in poverty. While the dollar amount of actual cash transfers to poor families that year was $20.7 billion, the poverty gap the dollar amount needed to raise all poor families above the official poverty line was $58.5 billion, more than twice the amount provided. 4,8 Why Is the Poverty Rate So Much Higher for Children Than for Adults? To this point, the focus of the discussion has been on poverty among adults who live with children. If children were randomly distributed among all adults with children, then the child poverty rate would be the same as the poverty rate of adults with children. However, the rate of poverty among children 21.9% in 1992 was much higher than the 14.4% poverty rate among adults

7 Why So Many Children Are Poor 31 Table 2 Why Are Children Poor? Decomposition of the 1992 Child Poverty Rate Reasons for Child Poverty Percentage-Point Contribution to Child Poverty Rate 1. Children who live with adults who lack adult self-sufficiency Children who live with adults who are self-sufficient but lack the 4.6 additional income to also support their children, even counting means-tested cash benefits 3. Children who live in poverty because of the demographic 7.5 structure of their household 1992 Child Poverty Rate (sum of factors 1 through 3) 21.9 Source: Calculations by the authors from the March 1993 Current Population Survey. See Current Population Survey [CD- ROM], U.S. Bureau of the Census, Available from U.S. Department of Commerce, Bureau of the Census, P.O. Box , Atlanta, GA with children. This 7.5 percentage-point difference results because poor families have more children per adult than the population as a whole. Calculations based on the March 1993 Current Population Survey show that poor families have 1.61 children per adult, which is much higher than the 0.93 children per adult in nonpoor families with children. 9 The higher ratio in poor families is, in turn, attributable to two other characteristics of poor families. First, poor families with children have fewer adults than nonpoor families with children. For example, 55.7% of poor families with children have only one adult, while only 13.9% of nonpoor families with children have only one adult. The much discussed high rate of divorce and relatively high rate of nonmarital births both contribute to there being fewer adults in poor families with children. Second, poor families with children have more children on average (2.24 per family) as compared to nonpoor families with children (1.79 per family). Most children who live with either a nonmarried or a divorced parent do, in fact, have another parent who lives elsewhere. If the family were thought of as a combination of those two units, the rate of child poverty might be substantially lower. Thus, the higher poverty rate for children in one-parent families reflects demographic circumstances, not necessarily the economic circumstances of all the relevant adult family members. This distinction underlies the policy efforts in recent years to tie the economic wellbeing of children to the economic circumstances of both of their biological parents. Summary Table 2 summarizes the reasons for child poverty. Nearly 10% of children would be poor even if children were distributed uniformly among adults. The reasons for that amount of child poverty, then, are directly related to the reasons for adult poverty the skills and labor supply of workers and the market demands for labor of various types. Another 4.6% of children live in families in which the adults achieve self-sufficiency but do not earn or receive enough income to cover the additional needs of their children. An additional 7.5% of children are poor because they live in families with more children per adult than the average family. This particular breakdown of the child poverty rate is but one of the many ways one might partition and think about the causes of child poverty. Because this dichotomy distinguishes between how adults fare in earning sufficient income for themselves and their children on the one hand, and how adults structure their families in terms of the number of adults and number of children on the other hand, it suggests how different social policies may affect child poverty. Policies such as job training and education directed at increasing adults earnings

8 32 THE FUTURE OF CHILDREN SUMMER/FALL 1997 will increase their chances of providing for themselves and their children. Such policies can have a secondary salutary impact on child poverty in the long run if these moreeducated adults then choose to have fewer children. Of course, not all policies may have multiple impacts that reinforce one another. Consider the proposal to cap welfare benefits to a mother who has an additional child while already receiving assistance. If this family cap policy were successful in limiting the number of children born to the poor, it would lower the average number of children per adult in poor families and decrease child poverty by reducing the gap between adult and child poverty rates. However, it would also limit the effectiveness of cash transfers in aiding adults in meeting their family s needs. Unless adults met the increased needs of their families through increased earnings, adult poverty would rise, and correspondingly, so too would child poverty. Trends in Poverty for Children As shown in Figure 1, the child poverty rate has followed a U-shaped pattern over the past 35 years, and the gap between the child and adult poverty rates has been larger in recent years than in earlier years. The decline in poverty from the late 1950s into The official poverty measure does not accurately reflect the relative economic well-being of all families. the 1970s was, in large part, a product of the growth in the economy, which explains why that decline in poverty is seen for both children and adults. In the latter half of the 1960s, there was also a major effort through social legislation and cash transfer programs the war on poverty to alleviate poverty directly, and that program appears to have had an effect in those initial years. Since the early 1970s, the growth of the economy has slowed, and since the mid- 1970s, the real value of cash transfer programs for children (that is, AFDC) has declined, while retirement income and Social Security provisions for the elderly have continued to rise. Thus, social policy itself may be a primary reason for the growth in recent years in the poverty rate for children as compared with that for adults. When the overall economy is playing the key role, as in the early time period covered by Figure 1, the experiences of children and adults are similar; when social policy plays the central role in the pattern of poverty over time, then the rates for demographic groups can and often will differ. The relatively stronger commitment to addressing the needs of the elderly as compared with those of children in recent decades is part of the explanation for the rise in child poverty relative to adult poverty. However, it bears mentioning that the trends evident in Figure 1 are based on the official poverty measure, which, as is discussed in the next section, does not accurately reflect the relative economic wellbeing of all families. To the extent that social policy designed to aid children is not reflected in the official measure while policies to aid the elderly are, the trends in Figure 1 overstate the decline in children s well-being relative to that of adults. As discussed in the article by Currie in this journal issue, the proportion of aid to children that is given in kind (for example, the Special Supplemental Food Program for Women, Infants, and Children (WIC), school lunches, and Head Start) and, therefore, not counted in the official poverty measure s definition of a family s income relative to the proportion that is given as cash and counted in the official measure has risen steadily over the past 20 years. Thus, children have benefitted in ways not apparent in the official poverty measure and not apparent in the trends shown in Figure 1. A New Poverty Measure In the Family Support Act of 1988, Congress called for a scientific review of the official U.S. measure of poverty, reflecting the general dissatisfaction with that measure, which had not been revised since the mid-1960s. The National Research Council (NRC) of the National Academy of Sciences undertook that review through a panel of academics of which the authors were members. That panel s recommendations were made in May 1995 and called for a new approach to poverty measurement. 10 This section summarizes the NRC panel s arguments for

9 Why So Many Children Are Poor 33 changing the way poverty is measured in the United States and describes how the proposed changes affect the composition of the children who are in poverty. The NRC Panel s Critique and Recommendations The panel s critique of the official poverty measure emphasized that several judgments about the measure made in the 1960s are by now antiquated and erroneous, even if they had been satisfactory when they were introduced. For example, the panel identified several types of family resources that have increased since the mid-1960s but are currently excluded from income in the official measure. For instance, government transfer programs such as food stamps and public housing subsidies are excluded because the official poverty measure counts only monetary benefits like AFDC and does not consider in-kind benefits, despite the fact that they have become a key component of welfare policy. Many of these in-kind programs did not exist when the poverty measure was created, yet in recent years welfare policy has favored in-kind programs to provide assistance. In 1992, for example, total AFDC and SSI cash assistance was $47.7 billion, while inkind assistance programs totaled $56.2 billion. 11 To ignore these latter programs seriously overstates measured poverty. Another serious omission in the current measure of poverty, the NRC panel argued, is the practice of ignoring paid taxes. In the 1960s, the poor were practically exempted from federal income taxation, and very few states taxed low-income families; therefore, the only income-related tax paid by the poor was the Social Security payroll tax at 3% of earnings. Today, however, the poor are subject to considerably higher earnings taxes at both the federal and the state levels and also in Social Security taxes, which are currently at 7.56% of earnings. Failure to account for these higher taxes understates poverty. The panel argued that taxes should be taken into account by subtracting them from available resources. The official measure of poverty also ignores an important cash program administered through the tax code, the Earned Income Tax Credit (EITC). In 1992, this tax credit gave low-income households $9.6 billion of assistance, none of which was counted as income in the official measure of poverty. Ignoring this program, designed to aid working families with children, results in an overstatement of the number of poor children. In 1960, some 18.6% of married women with a child under six participated in the labor force; by 1994, this rate had more than tripled to 61.7%. The panel argued that this upward trend necessitates another change in the way poverty is measured. Certain child care costs should be subtracted from earnings available to meet poverty thresholds in families in which there are young children and no nonemployed adult at home. 12 Similarly, for each working adult, the panel recommended deducting a flat amount per week worked to account for work-related expenses such as transportation. To ignore these costs overstates available income and thereby understates the level of poverty. Another important trend in recent years that is not handled well in the official poverty measure is the rapid rise in health

10 34 THE FUTURE OF CHILDREN SUMMER/FALL 1997 care expenditures. Per capita medical spending in 1960 in real (1992) dollars was $698; by 1993, that figure had grown to $3,299. With a growing proportion of the population uninsured for health care, families today face an increased risk of medical spending. The NRC panel identified the issue of medical care spending as particularly problematic and recommended that the definition of family resources be altered to reflect the burden imposed by medical expenses. Another problem identified by the panel is that there is no recognition in the official poverty measure of differences in the cost of living across the United States, although the cost of housing in New York City, for example, is 162% higher than in rural Mississippi. To ignore the potentially Choice of a specific level for the poverty threshold cannot be determined by science alone and is appropriately determined by public debate and ultimately by public officials. large differences in the cost of living across regions and cities of different sizes tends to bias the measure of children in poverty toward identifying too many as impoverished in rural areas and in the South relative to children living in large metropolitan areas elsewhere. As indicated by this discussion, an improved measure of the income available to meet poverty thresholds would, in some instances, raise the number in poverty and, in other instances, lower that number. As income is currently measured, however, families are clearly ranked incorrectly by income, so the measure of which families are in poverty is flawed. Turning to threshold measurement, another fundamental issue addressed by the panel is how to update over time the dollar level of the poverty threshold, the line between poverty and sufficiency. If one thinks poverty should be defined by an absolute level of resources or needs, then updating poverty thresholds for changes in the cost of living using the Consumer Price Index, for example may be adequate. Alternatively, if one thinks poverty should be defined in comparison to the resources of an average family or a wealthy family, then the updating should be done in relation to the change in resources of that reference family. Although the United States is the only country with an official measure of poverty, many other nations informally use a relative measure for their poverty thresholds, often half the median income. The United Kingdom uses half the mean (not median) income as its poverty threshold; with this concept in particular, a family can become poor not because its income fell, but because another (even a very wealthy) family s income rose. There is no correct or incorrect answer to the question of whether to use an absolute or relative measure of poverty. However, the NRC panel noted the interesting pattern in Figure 2, which shows three potential measures of U.S. poverty thresholds over time, each expressed in real (1992) dollars: (1) the current poverty thresholds (an absolute measure of poverty); (2) one-half the median after-tax family income (a relative measure of poverty); and (3) the average response to a Gallup poll survey question, How much does a family of four need not to be considered poor? 13,14 Figure 2 shows that, in 1963, when the official (absolute) poverty threshold was set, there was consistency among the measures of what a family of four needed. But notice how unusual it was for those three measures to converge. Had our official (absolute) threshold been in place in the immediate post World War II period, it would have been considered far too high, and analogously by the subjective survey responses or the comparison to half-median income, it has been considered much too low in recent years. Although the official poverty thresholds are adjusted annually to reflect changes in prices, real threshold levels have never been changed, despite a nearly 30% increase in median after-tax incomes of four-person families since the mid-1960s when the thresholds were first introduced. The NRC panel argued for some upward adjustment in the poverty thresholds. Most important, it argued for a procedure that would update those thresholds over time as spending levels on necessities change.

11 Why So Many Children Are Poor 35 Figure 2 Absolute and Relative Measures of Poverty Thresholds for a Family of Four in the United States, 1947 to 1992 $20,000 $18,000 $16,000 $14,000 $12,000 $10,000 $8,000 $6, Annual Income (in 1992 dollars) Year Official poverty threshold a Subjective poverty threshold* Half of median after-tax family income a Subjective poverty threshold is the mean response to a national Gallup Poll survey question, How much does a family of four need not to be considered poor? Source: Citro, C., and Michael, R. Measuring poverty: A new approach. Washington, DC: National Academy Press, 1995, Tables 2-3 and 2-4. The choice of a specific level for the poverty threshold cannot be determined by science alone and is appropriately determined by public debate and ultimately by public officials. To narrow the debate, the panel recommended that any new poverty threshold should be based on how families actually spend their income on the necessities of food, clothing, and shelter, including utilities. Impact of the NRC Panel s New Measure of Poverty on Children For the purposes of this discussion, the threshold for a family of four is set so that the same number of children are counted as poor by the panel s concept of poverty as are poor under the current poverty measure. This exercise emphasizes the differences in which children are counted as poor, rather than emphasizing the impact that the choice of the threshold will have on the number of children who are counted as poor. This procedure reflects one of the key points made by the panel: setting the level of the threshold and, hence, the number of poor is principally a political judgment, but the ranking of families from the poor to the wealthier is less judgmental and can be based on scientific measurement. In 1992, the official poverty measure counted 14.6 million children as living in poverty. Holding constant that number of children but applying the panel s recommendations about the thresholds and the correct measure of the family s income level, it becomes apparent that as many as onesixth of the children counted as poor by this new measure were not poor by the official measure. That is, 2.4 million children who are counted as poor in the official measure

12 36 THE FUTURE OF CHILDREN SUMMER/FALL 1997 live in families with incomes that are higher than another 2.4 million children who are not counted as poor in the official measure but who are poor according to the panel s proposed measure. The explanation for this reranking is that the new measure counts as family resources in-kind benefits such as food stamps and income tax credits and subtracts from available family resources nondiscretionary expenditures like paid taxes. Among the children who are newly in poverty in this exercise, many more reside with two parents: according to the official measure, 36% of poor children reside with two parents, while in the new measure, 43% do so. (For comparison, in the population as a whole, 73% of children live with two parents.) Similarly, the percentage of poor children living with at least one full-time When employment-related expenses are accounted for, many working families no longer have adequate resources to meet family needs. employed parent increased from 26% under the official measure to 35% under the panel s measure. The explanation for both of these shifts is that, when employment-related expenses including child care and the taxes associated with the earnings are accounted for, many working families no longer have adequate resources to meet family needs. The racial and ethnic composition of poor children remains relatively unchanged through this exercise. While there are slightly more white children in poverty, black and Hispanic children remain overrepresented in the poverty population. Children who are classified as poor by the new measure are much less likely to be receiving a transfer payment from the government because, as one would hope, the receipt of benefits provides many families with resources that are sufficient to meet their needs. For those children who are no longer classified as poor, 71% lived in families that received AFDC or SSI, 87% lived in families that received food stamps, 43% lived in families that had housing subsidies, and 60% lived in families that received the EITC. The picture of the children reclassified as poor by the proposed measure is quite different. Five percent lived in families that received AFDC or SSI, 11% lived in families that received food stamps, 4% lived in families that received a subsidy for housing, but 74% of these children lived in families that received an EITC payment. As previously noted, the current poverty definition does not reflect any geographic differences in the cost of living. Adjusting the poverty thresholds to account for these differences creates a shift in the regional location of poor children from the South and the Midwest to the Northeast and the West. While the proportion of poor children living in the southern region declines, the South continues to have the highest proportion of poor children. Geographic cost-ofliving adjustments also reflect differences between rural and urban areas. After adjusting for these differences, there is a slight rise in the number of poor children in central cities and suburbs, with an offsetting decline in the proportion from rural areas. One final but important characteristic of poverty among children should be examined, the depth of child poverty. These measures focus on the amount of income inadequacy rather than on a count of the number of children who are poor. Figure 3 shows the proportion of children whose family s resources lie in the intervals less than 25%, 25% to 50%, 50% to 75%, or 75% to 100% of their family s needs as reflected by the poverty thresholds. Using the current, official measure of poverty, poor children are almost uniformly distributed among these four subpoverty categories. By contrast, using the NRC panel s measurement of poverty that takes into account the in-kind transfers and other adjustments, a much more optimistic picture emerges, with many more children in families closer to meeting their needs. Conclusion If the NRC panel s poverty measure were adopted, it would alter the statistical face of poor children, as compared with the current poverty measure. The revised measure provides a more accurate account of which children remain impoverished despite efforts by their parents and the government. With the

13 Why So Many Children Are Poor 37 Figure 3 Comparison of the Depth of Poverty Among Children Using the Current Poverty Measures and Those Proposed by the NRC Panel Percentage Current Poverty Measure NRC Panel Proposed Annual Family Resources as a Percentage of Poverty Threshold 0 to 25% 25% to 50% 50% to 75% 75% to 100% Source: Calculations by the authors from the March 1993 Current Population Survey. See Current Population Survey [CD-ROM], U.S. Bureau of the Census, Available from U.S. Department of Commerce, Bureau of the Census, P.O. Box , Atlanta, GA NRC panel s measure, poor children will continue to be disproportionately those living with a single parent and those who are black. In the panel s measure there is, however, a sizable increase in the portion of poor children who live in families where both parents are present, where at least one adult is working a full year, and where no governmental cash assistance is received. Many of these working, low-income, two-parent families are currently incorrectly not counted among the poor. In the first section of this article, the child poverty rate was decomposed emphasizing (1) the ability of parents to provide for themselves and for their children, and (2) the differences in the demographic structure of poor families compared with other families. It is instructive to look at that same decomposition using the panel s poverty measure. With that new measure, a larger proportion of the children in poverty live in households with two adults present. This implies a reduction in the number of children per adult in poor families with children and, hence, a narrowing of the gap between the poverty rates for adults and those for children. In fact, the average number of children per adult in poor families falls to 1.52 in the panel s poverty measure from 1.61 in the old measure, while the poverty rate of adults with children rises to 15.3% in the NRC s measure from 14.4% in the old measure. Also, with that new measure, a larger proportion of the children in poverty live with parents who are employed the full year. This implies that more adults should be able to achieve adult self-sufficiency. In fact, the proportion of adults with children who do not achieve adult self-sufficiency falls very slightly from 9.8% to 9.6% while the proportion of adults who fail to achieve family selfsufficiency rises to 5.7% from 4.6%. Finally, the new poverty measure takes account of the tax credits and noncash government assistance to families, which implies that the effect of these programs can now be measured. In the first section of this article, it was noted that only 0.9% of both adults and children were lifted from poverty

14 38 THE FUTURE OF CHILDREN SUMMER/FALL 1997 through means-tested cash assistance (AFDC and SSI). When the tax credits and in-kind benefits are also counted, 4.2% of adults with children and 5.3% of children are lifted out of poverty through governmental assistance. Clearly the old definition of poverty understates the contribution of the variety of governmental transfer programs to reducing child poverty. The United States has an official measure of poverty that has been utilized for more than 30 years and that has, ironically, become obsolete in part because of antipoverty initiatives. The NRC panel s recommended measure of poverty takes those initiatives into account and accommodates, as well, other social and economic changes that have taken place over the past three decades. As this nation moves toward devolution, with individual states initiating a wide variety of social welfare programs, it is essential that the national measure of poverty be sufficiently flexible to reflect the impact of those state policies. Some states may use cash payments, while others may use inkind transfers more extensively and still others may use vouchers or credits of some other kind. Unless the national measure of poverty accurately reflects the needs and the resources of families, there will be no way to determine which state s strategies work better and which work less well in alleviating poverty. 1. For an alternative view about the trend in child poverty over the past two decades, see Jencks, C., and Mayer, S.E. Do official poverty rates provide useful information about trends in children s economic welfare? Institute for Policy Research Working Paper, April 18, Lewit, E.M. Child indicators: Children in poverty. The Future of Children (Spring 1993) 3,1: One reason this is only a crude approximation is that the actual income has been taken by source that is, the earnings, the Social Security, and so on as if the presence or absence of children has no effect on these amounts, but in fact it surely does. There are important behavioral responses to the presence of children in terms of men and women s labor supply, the family s savings, and family stability and structure. The computations ignore these responses and simply count whether the income received in the year is sufficient to keep the adults living in the household out of poverty. 4. These calculations and all others in the first section of this article are based on calculations by the authors using data from the March 1993 Current Population Survey. See Current Population Survey (CD-ROM), U.S. Bureau of the Census, Available from U.S. Department of Commerce, Bureau of the Census, P.O. Box , Atlanta, GA There have been many excellent surveys of research findings on the determinants of earnings. These include Willis, R.J. Wage determinants: A survey and reinterpretation of human capital earnings functions. In Handbook of Labor Economics. O. Ashenfelter and R. Layard, eds. Amsterdam: North Holland, 1986, pp ; Levy, F., and Murnane, R.J. U.S. earnings levels and earnings inequality: A review of recent trends and proposed explanations. Journal of Economic Literature (September 1992) 30,3: ; and Blinder, A.S. The level and distribution of economic well-being. In The American Economy in Transition. M. Feldstein, ed. Chicago: University of Chicago Press for NBER, 1980, pp See Sawhill, I.V. Poverty in the U.S.: Why is it so persistent? Journal of Economic Literature (September 1988) 26,3: , and Lewit, E.M. Why is poverty increasing among children? The Future of Children. (Summer/Fall 1993) 3,2: Table 1 includes adults of all ages, both with and without children. Although the number of children ever born is negatively associated with the parent s educational attainment, in Table 1 that relationship is obscured by the inclusion of elderly adults who are both unlikely to have children present and quite likely to have a relatively low level of educational attainment. 8. This poverty gap is only a first approximation and would be a minimum amount because additional cash assistance would tend to lower the incentives of individuals to support themselves, so even more assistance might be needed to fill it. 9. To see why the difference between the adult poverty rate and the child poverty rate can be explained by the fact that poor families have more children per adult than the nonpoor population, consider the following example. The 14.4% poverty rate for adults with children conveys the information that 144 out of every 1,000 adults with children in the population will be

15 Why So Many Children Are Poor 39 poor. If every adult in the population has on average one child, then that 1,000 adults would represent 1,000 children as well. But if poor adults have, on average, say, 1.5 children, then the 144 poor adults would be living with 216 children, and the poverty rate for children would be about 40% higher than the poverty rate for adults. 10. Citro, C.F., and Michael, R.T., eds. Measuring poverty: A new approach. Washington, DC: National Academy Press, Programs included in this count were all food assistance programs, including food stamps and school breakfast and lunch programs, all housing programs such as Section 8 housing, and energy assistance programs. 12. The NRC panel recommended that these child care costs be capped at $2,400 for one child and $4,800 for two or more children. 13. The Gallup poll question was not asked after Denton R. Vaughan, Office of Research and Statistics, Social Security Administration, Washington, DC. Personal communication, February 10, Interpolations by the authors supplied missing data points for the following years: , 1965, 1968, 1972, , and